Tryptophan-2,3-Dioxygenase (TDO) deficiency is associated with subclinical neuroprotection in a mouse model of multiple sclerosis

The catabolism of tryptophan to immunosuppressive and neuroactive kynurenines is a key metabolic pathway regulating immune responses and neurotoxicity. The rate-limiting step is controlled by indoleamine-2,3-dioxygenase (IDO) and tryptophan-2,3-dioxygenase (TDO). IDO is expressed in antigen presenting cells during immune reactions, hepatic TDO regulates blood homeostasis of tryptophan and neuronal TDO influences neurogenesis. While the role of IDO has been described in multiple immunological settings, little is known about TDO’s effects on the immune system. TDO-deficiency is neuroprotective in C. elegans and Drosophila by increasing tryptophan and specific kynurenines. Here we have determined the role of TDO in autoimmunity and neurodegeneration in experimental autoimmune encephalomyelitis (EAE), a model of multiple sclerosis. We created reporter-TDO mice for in vivo imaging to show that hepatic but not CNS TDO expression is activated during EAE. TDO deficiency did not influence myelin-specific T cells, leukocyte infiltration into the CNS, demyelination and disease activity. TDO-deficiency protected from neuronal loss in the spinal cord but not in the optic nerves. While this protection did not translate to an improved overt clinical outcome, our data suggest that spatially distinct neuroprotection is conserved in mammals and support TDO as a potential target for treatment of diseases associated with neurodegeneration.


Results
Cloning and characterization of the reporter-TDO mouse. To determine the spatial and temporal regulation of TDO we generated reporter mice for TDO expression (reporter-TDO). We inserted a copGFP and a firefly luciferase gene into the TDO2 locus, starting at the start codon of the TDO2 gene (Fig. 1a). The construct was electroporated into embryonic stem cells of C57BL/6N mice. Accurate insertion of the transgene was assured by southern blot analysis (Fig. 1b). A stop codon at the end of the reporter construct prevents TDO2 transcription. Western blot analysis shows that homozygous reporter-TDO mice are deficient in hepatic TDO (Fig. 1c, Supplementary Fig. S1). Hemizygous and homozygous reporter-TDO mice do not show any phenotype in organogenesis and breeding behavior. HPLC measurements revealed that serum trp and kyn levels are comparable to previously described TDO-deficient mice (TDO− /− ) 26 : Trp levels are elevated 8.8-fold in TDO− /− mice and 5.1-fold in homozygous reporter-TDO mice (Fig. 1d), kyn levels 2.2-and 1.9-fold, respectively (Fig. 1e). Elevated kyn levels in TDO− /− mice have been described before and it has been hypothesized that peripheral IDO accounts for this elevation 26 . Indeed, qPCR analysis shows that IDO expression is elevated in spleens but not in lymph nodes of TDO− /− mice ( Supplementary Fig. S2a,b). Hepatic IDO expression is not altered in TDO− /− mice and expression levels are by two orders of magnitude lower than in lymphoid organs ( Supplementary Fig. S2c). The above hypothesis is further substantiated by our measurements of trp and kyn in IDO-deficient mice (IDO− /− ) (Fig. 1d,e): while serum trp levels are similar in WT and IDO− /− mice, kyn levels are below detection limits of our HPLC set-up (0.5 uM), indicating that hepatic TDO maintains systemic trp levels and hepatic kyn is rapidly further metabolized, while systemic kyn levels are produced by IDO.
Hepatic TDO is induced during inflammation. We detected hepatic luciferase activity in reporter-TDO mice, which increased during the course of EAE (Fig. 2a,b). This was confirmed on the transcriptional level by qPCR analysis from liver tissue of immunized WT mice at the indicated time points (Fig. 2c). The effect is not MOG35-55 specific but a response to the induced inflammation, as reporter-TDO mice immunized with Complete Freund's Adjuvant (CFA) alone emit similar levels of bioluminescence ( Supplementary Fig. S3a,b). It is also not a consequence of weight loss and the integrated stress response (ISR) as starvation for three days, a well-known inducer of the ISR, did not enhance but rather suppress TDO expression ( Supplementary Fig. S3c,d).
No luciferase could be detected in the CNS, spleens and lymph nodes of immunized mice (Fig. 2d). Of note, qPCR analysis of lymph nodes, spleens, brains, cerebella and spinal cords revealed that TDO2 is expressed in lymphoid and neuronal tissue, although at levels three to four orders of magnitude lower than in the liver ( Supplementary Fig. S4). No GFP expression was detected in the liver or CNS of the same mice, possibly due to posttranslational silencing of copGFP in the reporter construct.
The antigen-specific T cell response is unaltered in TDO-deficient mice. When comparing in vitro T cell proliferation from immunized WT and TDO− /− mice, no significant differences were detected (Fig. 3a). ELISA measurements revealed that cytokine profiles of T cells isolated from draining lymph nodes of immunized mice were not altered by TDO deficiency (Supplementary Fig. S5). These findings were substantiated by intracellular flow cytometry of T cells from lymph nodes of WT and TDO− /− mice (Fig. 3b,c) and CNS infiltrating CD4+ T cells (Fig. 3d,e), which did not show any shift in Th cell differentiation. This indicates that TDO deficiency does not impact the pathogenic T cell response during neuroinflammation. TDO does not affect immune cell infiltration and glial activation in the CNS. We next assessed whether TDO influences immune cell infiltration and glial activation in the CNS, both hallmarks of MS-and EAE pathology. Immunohistochemistry revealed no significant difference between WT and TDO− /− animals with respect to inflammatory lesion load (Fig. 4a,b), infiltrating CD3+ T cells (Fig. 4c,d) (Fig. 4e,f). Also glial activation remained unchanged, as shown by immunohistochemistry for the astroglial marker GFAP (Fig. 4g,h) and the microglial marker Iba1 (Fig. 4i,j).

TDO does not change demyelination and clinical disease activity in EAE.
In line with these data, myelin degradation remained unchanged in TDO-deficient mice, as determined by Luxol-PAS staining (Fig. 5a,b). Accordingly, there were no differences in clinical disease activity between WT and TDO− /− mice as measured by the degree of paralysis (Fig. 5c). Even when using a weaker immunization protocol resulting in a later onset and a milder disease course, the clinical disease activity was not aggravated by TDO-deficiency (Fig. 5d).
Neuroprotective vs. neurotoxic effect of TDO. As some kynurenines elicit excitotoxic and others neuroprotective effects, we analyzed signs of axonal degeneration. Immunohistochemistry of inflamed spinal cords revealed reduced numbers of β -amyloid precursor protein (APP) positive axons in TDO− /− mice ( Fig. 6a,b), which accumulates in the terminal axons during Wallerian degeneration. This indicates increased neuronal survival in TDO− /− mice, despite similar levels of inflammation and demyelination as well as degree of paralysis. We confirmed this subtle neuroprotective effect with immunohistochemistry of spinal cord slices for non-phosphorylated neurofilament (SMI-32), a stress marker for myelinated axons 30 (Fig. 6c,d).
To analyze the spatial specificity of this effect, we looked at the optic nerves, a second distinct neuronal compartment frequently affected in MS. Here, immunohistochemistry for APP and Bielschowsky's silver impregnation did not show any differences between TDO-deficient and WT mice ( Fig. 6e-h). Interestingly, staining of the retinal ganglion cell marker (RGC) Brn3a revealed reduced numbers of RCGs in TDO− /− EAE mice (Fig. 6i,j). Functional testing of the optic tract using visual evoked potentials (VEPs), however, did not reveal relevant differences (Fig. 6k,l). Collectively, these data indicate that TDO controls spatially distinct neurodegeneration in autoimmune neuroinflammation, albeit at a subclinical level.

Discussion
IDO-mediated trp metabolism suppresses antigen-specific T cell immunity and innate immunity in models of allergy, asthma, infection, cancer and autoimmunity [31][32][33] including EAE [15][16][17] , where natural and synthetic kynurenines modulate antigen-specific T cell responses 34 . Recently, we and others have shown that TDO-driven tryptophan catabolism in glioblastomas is equally capable of suppressing antitumor immune responses 10,11 . In addition, several kynurenines are well-described neuroactive metabolites. As TDO expression has been observed in distinct neuronal compartments it was tempting to hypothesize that TDO might impact immune responses and neuronal survival in a model of autoimmune neuroinflammation.
Our study shows increased hepatic expression and activity of TDO during EAE. However, our data do not support a role of TDO in modulating antigen-specific T cell responses in the EAE model. The discrepancy between IDO and TDO can be explained by different mechanisms. Most obvious are the divergent expression profiles of the two enzymes: In vivo bioluminescence analyses of reporter-TDO mice indicate that TDO is mainly expressed in the liver while the signal strength was not sufficient to detect TDO expression in lymphoid and neuronal tissue, which is by three to four orders of magnitude lower than the hepatic expression levels. It has been shown that within the CNS TDO expression is restricted to distinct cell subsets such as hippocampal neurons 24,25 (http://mouse.brain-map.org/experiment/show/70565321). In contrast, IDO is expressed ubiquitously and strongly induced in APCs by proinflammatory mediators and it changes trp and kyn levels in the local micromileu. IDO is highly expressed in the induction phase of EAE not only in secondary lymphatic organs but also in the spinal cord and several areas of the brain 16,17 . The divergent roles of IDO and TDO are supported by our HPLC measurements of trp and kyn serum levels in TDO− /− and IDO− /− mice: hepatic TDO is the main trp metabolizing enzyme which maintains systemic trp levels. Hepatic kyn is very likely rapidly metabolized. IDO in turn is responsible for producing kyn in the peripheral tissue and thereby for serum kyn levels. Its mRNA expression is elevated in spleens of TDO− /− mice. IDO tightly controls kynurenine levels in the local microenvironment, but it does not change systemic trp levels. This is supported by previous studies which found that IDO inhibition and IDO deficiency exacerbated EAE regardless of kyn serum levels 15 .
IDO accumulates kyn but depletes trp in the extracellular compartment, which sustains two synergistic mechanisms: Trp depletion leads to immunosuppression via the initiation of the amino acid starvation response by GCN2, which senses amino acid deficiency through uncharged tRNA. Its effects on immune reactions is best studied in CD4+ and CD8+ T cells where trp depletion creates T cell anergy 5 . Accordingly, chronic neuroinflammation is exacerbated in GCN2-deficient mice 35,36 . At the same time accumulation of kyn and downstream metabolites in the local microenvironment leads to immunosuppression by activating the AHR, an immunoregulatory receptor which is best studied on T cells in EAE and tumor models [6][7][8]10,37 . In contrast, hepatic TDO deficiency raises both, serum levels of trp and kyn in the serum, possibly resulting in two opposing immunoregulatory mechanisms.
Kynurenic acid is a well known neuroprotective factor while other kynurenines like 3-hydroxykynurenine and quinolinic acid are excitotoxins 29,38-41 . It has been shown that TDO deficiency elevates merely all kynurenines in serum and brain tissue as demonstrated for kyn, kynurenic acid, 5-hydroxyindoleacetic acid, indolelactic acid and indoleacetic acid 26 . TDO-deficient mice display a mild phenotype with subtle alterations in anxiety-related behavior 26 . Recent studies in flies and lower vertebrates confirmed the neuroactive role of TDO: TDO inhibition and accumulation of kynurenic acid is neuroprotective in a Drosophila model of Huntington's and Alzheimer's disease 27,29 and in C. elegans TDO-deficiency extends the animal's lifespan by suppressing proteotoxicity 28 . Here, it also appears that the systemic rather than the local CNS dysbalance of trp metabolites is responsible for the phenotype.
Our results extend these observations to the mammalian system and support the concept that increasing systemic trp and its metabolites result in a net neuroprotective phenotype. In addition to our observation that this phenotype in the EAE model is subtle and at a subclinical level, we found distinct effects on neurons of different systems: While there is a protective effect of TDO deficiency in spinal axons, this finding was not recapitulated in neurons of the visual system, rather the contrary, we saw a slight but significant additional neurotoxic effect on Brn3a+ RGC in TDO− /− mice.
This divergent effect could be explained by i) differential susceptibility to neurotoxicity and ii) differential expression of TDO in certain areas of the brain. Differential susceptibility to excitotoxicity induced by the trp metabolites quinolinic acid and 3-hydroxykynurenine is mainly determined by distinct expression patterns of NMDA receptor subunits 42 . Indeed, RGC express functional NMDA receptors 43 , and while early reports described their relative resistance to excitotoxins 44 , more recent research showed that this resistance applies mainly to the subset of OPN4+ intrinsically photosensitive RGCs, whereas Brn3a+ traditional RGCs are susceptible to excitotoxicity by NMDA 45 . Spinal neurons, while expressing NMDA receptors and being responsive to glutamate, show only low excitability to quinolinic acid 46 , and therefore the protective effects of kynurenine and kynurenic acid may prevail. In contrast, hippocampal and striatal neurons are excitable by quinolinic acid 46 , (while at the same time expressing TDO, see below). In addition to excitotoxicity via direct NMDA receptor activation, quinolinic acid can be toxic to neurons and glial cells (i) by increasing microenvironment glutamate concentration via inhibiting glutamate release by neurons and its uptake by astrocytes, (ii) by potentiating the effect and toxicity of other NMDA receptor antagonists, and iii) by formation of reactive oxygen species which cause lipid peroxidation (for a comprehensive review, see ref. 47). TDO expression varies highly between different neurons with by far the highest expression in granule cells of the hippocampus 24,25 (http://mouse.brain-map.org/experiment/show/70565321). Inflammatory lesions, demyelination and neurodegeneration in our EAE model are mainly restricted to the spinal cord and the cerebellum 48 . In human MS however, a majority of inflammatory lesions are located in paraventricular white matter, marked hippocampal atrophy is characteristic to the disease 49 and cognitive impairment is present already at early stages of MS 50 . Therefore, TDO inhibition could prove to be more relevant in human MS than what we might deduce from its mouse model.
In summary, our study suggests that suppressing systemic TDO activity -as previously suggested in worms and flies -may represent a viable neuroprotective strategy in diseases associated with neurodegeneration such as multiple sclerosis.  control animals. Transgenic reporter mice for TDO (B6-Tdo2tm#(GFP/Luc)/Dkfz, here referred to as "reporter-TDO") were created on a C57BL/6N background with JM8A1.N3 embryonic stem (ES) cells. All used animals were female and 8-12 weeks old. Animals were housed in the SPF animal facility of the German Cancer Research Center (DKFZ) in Heidelberg, Germany. All animal work was performed in accordance with the European and German animal protection law with approval from the "Regierungspräsidium" in Karlsruhe, Germany and the "Tierlaborausschuss" of the German Cancer Research Center (DKFZ) in Heidelberg, Germany.
Cloning of the transgenic reporter-TDO mouse. The transgenic construct containing a copGFP and a firefly luciferase gene in the TDO2 locus, starting at its start codon (Fig. 1a), was cloned using an artificial chromosome (BAC) and Red/ET technology 52,53 . A stop codon was inserted at the 5′ end of the insert, therefore a homozygous mouse results in a knockout of TDO. Red/ET cloning was carried out by GeneBridges, Heidelberg, Germany. The BAC RP23-36B19 was subjected to PCR analysis for identity and integrity and cloned into a high-copy plasmid backbone, then it was again checked by restriction digest. In a first step copGFP/firefly luciferase was inserted into the plasmid by multi-site Red/ET recombination. In a second step, a kanamycin/neomycin resistance cassette was included again by multi-site Red/ET recombination. As only very few plasmid copies in a cell undergo Red/ET recombination very little DNA of a mosaic positive clone was used to transform new cells. A positive clone was identified by restriction digest and sequenced. The plasmid was linearized using the restriction enzymes SalI and SacII before electroporation into ES cells.
Embryonic stem cell culture. 2 × 10 6 JM8A1.N3 ES cells were cultured on inactivated feeder cells plated on the day before. After 5 days of culture cells were trypsinized and electroporated with 50 μ g of the DNA construct using a gene pulser electroporation system (BioRad, Hercules, CA, USA). After another 24 h of cell culture ES cells selection started by addition of gentamycin (G-418, Sigma-Aldrich, Seelze, Germany). After 7 days of culture under selection conditions, appr. 350 clones were picked and further cultured in 48 well plates for two more days under selection. Then cells were trypsinized and 2/3 were frozen for later blastocyst injection while 1/3 was cultured on gelatinized 24-well plates for another 6-10 days for southern blot analysis.
Blastocyst injection. Superovulation of 4-8 week old female mice was achieved by i.p. injection of 7 IU pregnant mare's serum gonadotropin and 46-48 h later by 7 IU human chorionic gonadotropin. 1 h after the last injection mice were fertilized by congenic males and checked for a vaginal plug. Blastocysts were obtained at day 3.5 by flushing of the isolated oviduct/uterus. Transgenically altered embryonic stem cells were injected into the blastocoel. Blastocysts were injected into the uterus of pseudopregnant foster mothers (after mating with vasectomized males). Genotyping of the first transgenic/mosaic litter was achieved by fur color. Positive mice were bread with WT C57BL/6N mice and germline transmission was screened by PCR and southern blotting. Transgenic mice were crossed to Flp-deleter mice (B6.CG-TG(ACTFLPe)9205Dym) to erase the neomycin cassette. Correct insertion of the reporter cassette was checked by southern blotting before and after Flp deletion (Fig. 1b). Homozygous females were used for bioluminescence imaging.
Southern blot analysis and PCR genotyping. Southern blot strategy including the expected band sizes is depicted in Fig. 1a. Southern blot analysis was achieved according to standard protocols using a DIG DNA Labeling and Detection Kit (Roche, Basel, Switzerland). Briefly, a 1049 bp template DNA was amplified using the primers: forward: 5′ -GGAAGGGGTGGAAACCAGAGGC-3′ and reverse: 5′ -AGAGGAAGTTGTTCGGACGCTGC-3′ . The template was labeled with Digoxigenin-11-dUTP. Genomic mouse DNA was isolated from embryonic stem cells or tails by incubation at 56 °C for 2 h with 1 μ g/ml Proteinase K (Sigma) in DNA lysis buffer, containing 100 mM Tris-HCl, 200 mM NaCl, 5 mM EDTA and 0.2% SDS, followed by an inactivation step at 95 °C for 20 min. Genomic DNA was precipitated with isopropanol and washed with alcohol, then 10 μ g were digested using BspH1 (New England Biolabs, Ipswich, MA, USA) in NEB Buffer 4 overnight before electrophoresis on a 1% agarose gel. DNA was denaturated on the gel by incubation in 0.5 M NaOH/1 M NaCl for 0.5 h, rinsed and neutralized twice with 0.5 M Tris, pH7. 4/3 M NaCl for 20 min. The gel was placed in 20× saline-sodium citrate buffer with a nitrocellulose membrane on top. DNA transfer onto the membrane was achieved by capillary transfer overnight. The membrane was dryed at 80 °C for 2 h and hybridized with the DIG-labeled template in 4 × SSPE, 5× Denhardt's reagent, 1% SDS, 100 μ g/ml salmon sperm DNA and 1% blocking reagent at 65 °C. The membrane was washed 3× , blocked for 0.5 h with blocking reagent, incubated with anti-DIG solution (1:5000), then washed and incubated for several hours in NBT and Bcip until DNA staining became visible.
Monitoring of phenotype. The following parameters have been monitored in homozygous reporter-TDO mice in particular: breeding behavior, litter size, litter per year, surviving litter size as well as weight and size after birth until week 4. Macroscopic autopsy has been performed on 5 adult mice at the age of 10-12 weeks with a special focus on abdominal organs (liver, spleen, stomach, intestine, kidneys, pancreas, bladder) as well as lung, heart and brain, and they showed no abnormalities in appearence, shape and weight. As TDO is mainly expressed in the liver, we examined hepatic H&E stainings, also showing no difference to livers from WT mice.
Western blot analysis. Whole-liver lysates were prepared with high-speed ball mill (Tissue Lyser II T cell Proliferation and Cytokine Measurement. For the measurement of antigen-specific proliferation of T cells, mice were sacrificed on day 9 after immunization with MOG p35-55. Cells were mechanically singularized as described above and processed as described earlier 4 . Briefly, lymph node cells were seeded in U-bottom 96-well plates at 5 × 10 5 cells per well in enriched RPMI, antigen specific activation was achieved by addition of 5-20 μ g/mL MOG p35-55. For measurement of T cell proliferation, cells were pulsed after 72 h with 3H-methylthymidine (Amersham-Pharmacia Biotech, Amersham, UK) at 1 μ Ci per well for the last 18 h. Cells were harvested using a Tritium Harvester (Tomtec, Unterschleissheim, Germany) and a β -plate reader (Wallac/ PerkinElmer, Wellesley, MA, USA) with BetaWin software. For cytokine measurements, supernatants were taken after 72 h of culture and measured by ELISA for the cytokines IL-2, IL-4, IL-6, IL-17A, and IFN-γ (ebioscience) and after 96 h for the measurement of IL-10 (ebioscience) according to the manufacturer's instructions.
Tissue preparation and immunohistochemistry of spinal cords. Mice were anesthetized by i.p.

Histopathological analyses of spinal cords. Statistical evaluation of infiltrating immune cells was
obtained from H&E stainings by manually counting lesions of infiltrating cells per slide. Demyelination was evaluated by manually counting demyelinated areas in luxol-PAS-stained spinal-cord slides, correlating it to the following score: 0.5, single demyelinated spot; 1, several spots; 2, confluent sites of demyelination; 3, demyelination of one half of the spinal cord; and 4, demyelination of more than half of the spinal cord 55 . GFAP and Iba1 immunohistochemistry was manually evaluated by counting the number of single stained cells with a corresponding nucleus per spinal cord slide. Immunohistochemistry images for APP and non-phosphorylated neurofilament (SMI-32) were manually evaluated by counting the number of APP-positive axonal profiles per spinal cord slide. CD3 and Mac3 immunohistochemistry was evaluated using Fiji/ImageJ software 56 , correlating the stained area to the whole spinal cord area. Morphological analyses and manual counting were carried out by a blinded investigator. All statistical evaluations were carried out in multiple slides of multiple animals as indicated in the corresponding figure legend.
Optic nerve histopathology. Healthy mice and those at day 21 of EAEwere anesthetized with 200 mg/kg ketamin and 40 mg/kg xylazine i.p. and transcardially perfused with 4% paraformaldehyde in PBS. Optic nerves were dissected, processed for paraffin-embedding and 0.5 μ m transverse sections were cut. Bielschowsky's silver impregnation was performed in order to depict axonal pathology, as previously described 55 . To further assess axonal degeneration, immunohistochemistry was performed using an antibody against APP as described previously 57 . Retinal histopathology. Following perfusion (see above), eyes were enucleated, post-fixed in 4% paraformaldehyde for 2 hours and cryoprotected in 30% sucrose overnight. Eyes were then embedded in mounting medium (Tissue-Tek O.C.T Compound, Sakura Finetek Europe, Alphen aan den Rijn, Netherlands) and frozen. Twelve μ m coronal sections were sliced for immunofluorescence. Sections were washed in PBS and blocked in 10% sera in 0.1% PBS-Triton X. Sections were incubated with anti Brn3a (Santa Cruz Biotechnology Inc, Santa Cruz, CA, USA) and staining was visualized with the an Alexa 555-conjugated secondary antibody (Jackson Immuno Research Laboratories, West Grove, PA). Sections were mounted in anti-fade medium containing 4′ ,6-Diamidin-2-phenylindol (DAPI) (Vector Laboratories).

Histopathological analyses of optic nerves and retinae.
For all histopathological investigations, a minimum of 10 sections taken throughout either the length of each optic nerve or the retina were quantified. The number of APP positive axons was quantified using a morphometric grid and calculated per mm 2 of the optic nerve. Relative axonal densities were determined in optic nerves cross sections stained with Bielschowsky's silver impregnation by point sampling using a 25-point Olympus eyepiece, as previously described 57 . The degree of axon reduction is given as the percentage of axon density compared to the average axon density in healthy optic nerves. The number of Brn3a positive cells was quantified using a morphometric grid and calculated per mm of the retina.
Visual evoked potential (VEP) measurement. Two weeks prior to immunization, and again at day 21 of EAE, animals were anaesthetised i.p. with ketamine (120 mg/kg) and xylazine (10 mg/kg). Measurements of visual evoked potentials (VEPs) were performed using the UTAS Visual Diagnostic System (LKC Technologies, Gaithersburg, MD, USA). Mice were placed on a heated pad and the temperature maintained at 37 °C. Prior to recording, pupils were dilated with 0.5% atropine (Ursapharm, Saarbrücken, Germany) and the animal dark-adapted for 5 minutes. Needle-type electrodes were placed in the primary visual cortex, 3mm lateral to lambda. A reference electrode was placed subcutaneously in the neck and the ground electrode was placed Scientific RepoRts | 7:41271 | DOI: 10.1038/srep41271 subcutaneously in the tail. Animals were placed in the dome equipped with a LED wholefield stimulator. Flash stimuli were presented from a distance of 20 cm with 0 dB intensity and frequency of 2 Hz. 100 sweeps were averaged for each recording. During the recording process, desiccation of eyes was prevented through the use of Liquifilm ® O.K. eye drops (Allergan, Westport, Ireland). This recording procedure was performed on each eye separately, and again on day 21 of EAE. The signal amplitude (μ V) and latency (ms) were calculated from the first negative (N1) to the second positive (P2) peak of the response using software provided by the UTAS Visual Diagnostic System.